*June* 2020 - HackMD

# *June* 2020 ###### tags: `2020 dev call` Call co-ordinates: Mondays @ 15:00 Eastern US time (UTC -05:00) https://zoom.us/j/384435716 Previous notes: [2020 Meeting Agenda](https://hackmd.io/zljR-pZrQ0O5J_j4NZ-9yw) Archive: [matplotlib/ProjectManagement](https://github.com/matplotlib/ProjectManagement) #### [Needs discussion at some point](https://hackmd.io/uzWviu8zSUChq3XhI2FqPg) # 29 June 2020 - [x] GH sponsors page https://discourse.matplotlib.org/t/github-sponsors/21320 / https://github.com/sponsors/matplotlib - [ ] named roles - [x] 3.3.0rc2 or 3.3.0 ? - [X] [#16090](https://github.com/matplotlib/matplotlib/pull/16090) pcolor snapping - [ ] https://github.com/matplotlib/matplotlib/pull/17265 organization of API changes ## Notes: ### GH Sponsors page * tiers * $1 dollar is important * Notes: * strip out promises * add more recognition especially at higher teirs * clean up copy * contract folks * fix mpl cairo ### 3.3.0 final * what's new needs to be cleaned up * 17764 * 17772 ### GSOC * one commit moving images from one directory to other directory * second commit is the code ### [17265:Mnt rearrange next api again](https://github.com/matplotlib/matplotlib/pull/17265) * use towncrier: https://towncrier.readthedocs.io/en/actual-freaking-docs/ ### drop 3.6 # 22 June 2020 ## Agenda - [x] macstadium access - [x] RaspberryShake arrived - [ ] GSOD selection committee - [X] rcparams for concise date formatter (@jklymak) - https://github.com/matplotlib/matplotlib/pull/17022 - [ ] parallel coordinates plot (PCP) ## Notes ### Macstadium access - macmini available. Ask @tacaswell, @annzter, @QuLogic for account. ### GSOD - Tim volunteered (under suggestion) ### concise date formatter rcParams * AxisInfo - static method to normal class method * is static everywhere * add image demoing new style ### Parallel coordinates plot - https://en.wikipedia.org/wiki/Parallel_coordinates - use axes as canvas w/ multiple spines and locators and stuff - add as an example and if it's useful/folks use it a lot and the example is complicated - add it in as a function - create new first issue w/ medium difficulty ### 3.3 issues - https://github.com/matplotlib/matplotlib/issues/17703 - call `autoscale_None()` aggressively - Caswell will write up series of example to make sure we all understand the previous and current behavior ### 3.4 changes - https://github.com/matplotlib/matplotlib/pull/17723 # 15 June 2020 ## Agenda - 3.3 / 3.2.2 - GSOC ## Notes - https://github.com/matplotlib/matplotlib/issues/16910 / https://github.com/matplotlib/matplotlib/pull/17636 - adjust temporarily adjust the vmin/vmax ### 3.3 / 3.2.2 - https://github.com/matplotlib/matplotlib/pull/17223 - Warn on invalid savefig keyword arguments - approved - https://github.com/matplotlib/matplotlib/pull/17635 - Fix autoscaling with tiny sticky values - approved - some concern about the slop in comparison being a general problem (this is the second one we talked about today) - should open an issue about cleaning up streamplot to avoid needing the slop in the sticky edges. - https://github.com/matplotlib/matplotlib/pull/17564 - FIX: correctly handle large arcs - @mdboom: > the code uses a Bezier approximation of an arc because that's way more efficient for formats like Postscript that didn't have native arc support. When the arc is so large that the approximation would be visible, it switches to a mode where it subdivides the arc into multiple splines. That inv_error magic number comes the paper linked in the docstring there -- I have to admit I took it on faith and didn't reverse engineer that value, however. - likely still something not quite correct about how we estimate the size and if we need to go the fast or slow path, but it is more correct than it was. - https://github.com/matplotlib/matplotlib/pull/17600 - FIX: work with PyQt 5.15 - change to int to be consistent, at some point make every int/round bit consistent w/ some needs to be created docs - https://github.com/matplotlib/matplotlib/pull/15656 - Support fractional HiDpi scaling with Qt backends - https://github.com/matplotlib/matplotlib/pull/17625 - Give `_DummyAxis` instances a `__name__` - minor suggestion, EFiring can self merge after call - https://github.com/matplotlib/matplotlib/pull/17471 - Removed idiomatic constructs from interactive figures docs - Caswell/Aizenman will review / rebase / merge after call # 8 June 2020 ## Agenda - [ ] 3.3 / 3.2.2 status ## notes ### 3.3 / 3.2.2 - https://github.com/matplotlib/matplotlib/pull/17580/files - consensus is to not backport - https://github.com/matplotlib/matplotlib/pull/17564 - provide image to make changes easier to review - https://github.com/matplotlib/matplotlib/pull/17560 - will add helper method/function to clean up inheritence - https://github.com/matplotlib/matplotlib/pull/17086 - remove jquery & use vanilla javascript for interactions - tests for webguis would maybe be welcome - https://github.com/matplotlib/matplotlib/pull/16603 - axes collage: - `subplots_compose` - `subplots_mosaic` - https://github.com/matplotlib/matplotlib/pull/17471 - waiting on tom - https://github.com/matplotlib/matplotlib/pull/17163 - should merge - https://github.com/matplotlib/matplotlib/pull/15532 - pushed to 3.4, have concerns - https://github.com/matplotlib/matplotlib/pull/15436 - pushed to 3.4, have concerns about implemenation - https://github.com/matplotlib/matplotlib/pull/15826 - Elliott will push a test - # 1 June 2020 ## Agenda - [x] 3.3 work? - https://github.com/matplotlib/matplotlib/pull/17515 - [x] 3.2.2 work? - [ ] Constrained layout rewrite - https://github.com/matplotlib/matplotlib/pull/17335 (easy) - https://github.com/matplotlib/matplotlib/pull/17347 (hard) - [ ]FuncNorm: * [#7631](https://github.com/matplotlib/matplotlib/pull/7631) & [#7294](https://github.com/matplotlib/matplotlib/pull/7294) & [#16391](https://github.com/matplotlib/matplotlib/pull/16391) * [#16457](https://github.com/matplotlib/matplotlib/pull/16457) * [x] Axes Collage * https://github.com/matplotlib/matplotlib/pull/16603 * actually release critical? * documentation strategy ## Notes ### Topic #### brief discussion of cases constrained layout has issues with - Can get it under-specified by making columns / rows that can be any size and still fulfill the constraints ### 16603: Axes Collage - new section called experimental in tutorials - provisional or experimental - name? - `subplot_grid` -> conflicts w/ `subplot2grid` - `subplot2grid` in`pyplot` namespace - can use fig.add_gridspec - `subplot_mosaic` - discoverability? - in docs make sure to have keywords: `patchwork`, `axes composition`, 'layout managers', `` - between subplot_mosaic and compose_subplots ### 17515: plot3d & bbox_inches='tight' - needs another reviewer - tightbox temporarly decoupling transfroms from size figure thinks it is - aspect 'auto' says don't adjust itself, but now aspect adjust itselfs - on axes 3d it's monkeypatched - aspect on 3d can be deleted - set_aspect - set_box_aspect added for ratio tuples - is tight_bboxing even done correctly? - longer term effort, leaves a big foot cannon laying around ### 17470: TK guardrails - too big images crashes xorg - add warning in figure about too large image if x or y is > 1000? - tends to be intentional ### FuncNorm - * [#7631](https://github.com/matplotlib/matplotlib/pull/7631) & [#7294](https://github.com/matplotlib/matplotlib/pull/7294) - has a specification language - [#16391](https://github.com/matplotlib/matplotlib/pull/16391) - issue was in application to symlog - break out FuncNorm into its own PR - basic idea is FuncNorm should be a two callabble, (forward/reverse) similar to FuncScale - [16457](https://github.com/matplotlib/matplotlib/pull/16457) - lognorm/symlognorm from scale - build norm from scale, match signatures - every norm has a scale private attribute which can be used by the colorbar - FuncNorm could forward to FuncScale ### no more jquery - https://github.com/matplotlib/matplotlib/pull/17086 ### [#17534: extended boundary norm](https://github.com/matplotlib/matplotlib/pull/17534) * added extra validation * don't specify more colors than bins